← 返回 Skills 市场
Secrets Scanner
作者
Anmol Nagpal
· GitHub ↗
· v1.0.0
463
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install secrets-scanner
功能描述
Detect hardcoded secrets, exposed API keys, and credential misconfigurations in IaC and config files
安全使用建议
This skill appears to implement the stated goal (finding secrets) but has ambiguous instructions that could lead to accidental leakage of real secrets. Before installing or using it:
- Never paste real secret values. Only paste redacted files (replace secrets with <REDACTED>) or provide filenames/paths for local scanning.
- Ask the skill author to clarify and fix the contradictory guidance: if the tool needs only environment variable names, the SKILL.md should show safe AWS CLI queries that return only keys (not values), or explicit jq/JMESPath examples to strip values before copy/paste. For example, prefer commands that output only keys or run local commands like: aws lambda get-function-configuration --function-name NAME --query 'keys(Environment.Variables)'. (Verify the JMESPath expression locally.)
- Confirm why ssm:DescribeParameters is included in the minimal IAM permissions and whether any additional permissions are required.
- Prefer running any automated scanning locally (open-source tools like git-secrets, truffleHog, or local regex/high-entropy checks) rather than pasting data into a remote assistant session.
- If you must share data with the skill, redact secret values first and confirm the skill's explicit rule to never output actual secret values.
If the author can remove the contradictory CLI examples or explicitly show/how to redact values and justify the ssm permission, the skill becomes much less risky and more coherent.
功能分析
Type: OpenClaw Skill
Name: secrets-scanner
Version: 1.0.0
The skill declares `bash` as a tool in `SKILL.md`, granting the agent shell execution capability. While the skill explicitly states it is 'instruction-only' and 'does not execute any AWS CLI commands,' the presence of `bash` introduces a significant vulnerability for potential Remote Code Execution (RCE) if the agent's internal guardrails are bypassed by a malicious prompt injection from a user. There is no evidence of malicious intent within the skill's own instructions, but the declared high-risk capability without a clearly defined, sandboxed use case makes it suspicious.
能力评估
Purpose & Capability
The name and description (detect hardcoded secrets in IaC/configs) align with the instructions which ask the user to provide files or CLI output. There are minor scope notes: the SKILL.md is AWS-focused (Lambda, ECS, Secrets Manager) so AWS-related permissions and recommendations are expected. However the included minimal IAM policy contains ssm:DescribeParameters which is not demonstrated in the example commands, and the metadata fields (price, pack: aws-security) are non-functional but consistent with a commercial scanner.
Instruction Scope
The instructions repeatedly say "do not ask for credentials" and to provide only exported or redacted data, which is good. But they also show sample aws CLI commands that will output environment variable names and values; elsewhere they ask for keys only. This contradiction is likely to lead users to paste sensitive values unintentionally. The guidance to "remove any actual secret values first" is necessary but relies on the user doing manual redaction. The skill explicitly says it won't run AWS CLI or access accounts directly (it is instruction-only), which is accurate and reduces autonomous risk, but the current instructions place the data-exposure burden entirely on the user.
Install Mechanism
This is an instruction-only skill with no install spec and no code files — the lowest install risk. Nothing is downloaded or written by the skill itself.
Credentials
No environment variables, binaries, or credentials are requested by the skill (primary credential: none). That is proportionate for a scanner that asks the user to paste/redact data. The only oddity is the provided minimal IAM permissions block (includes ssm:DescribeParameters) that isn't directly used in the shown commands; it should be clarified why SSM parameter access is listed.
Persistence & Privilege
The skill does not request persistent presence (always:false) and does not ask to modify agent-wide configuration. It is user-invocable and allows model invocation (default), which is normal for skills.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install secrets-scanner - 安装完成后，直接呼叫该 Skill 的名称或使用
/secrets-scanner触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
aws-secrets-scanner 1.0.0 - Initial Release
- Detects hardcoded secrets, API keys, and credential misconfigurations in IaC and config files.
- Instruction-only: analyzes exported data provided by the user; does not access AWS accounts or run AWS CLI commands directly.
- Supports scanning Terraform, CloudFormation, CDK, and extracted environment variable names from Lambda/ECS.
- Identifies various secret types including AWS keys, API tokens, SSH keys, connection strings, and hardcoded passwords.
- Produces actionable findings, risk assessment, and migration/remediation guidance (including AWS Secrets Manager integration and Git history cleanup).
- Protects sensitive data—never outputs raw credentials, only their locations and recommended next steps.
元数据
常见问题
Secrets Scanner 是什么？
Detect hardcoded secrets, exposed API keys, and credential misconfigurations in IaC and config files. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 463 次。
如何安装 Secrets Scanner？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install secrets-scanner」即可一键安装，无需额外配置。
Secrets Scanner 是免费的吗？
是的，Secrets Scanner 完全免费（开源免费），可自由下载、安装和使用。
Secrets Scanner 支持哪些平台？
Secrets Scanner 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Secrets Scanner？
由 Anmol Nagpal（@anmolnagpal）开发并维护，当前版本 v1.0.0。
推荐 Skills